Rosenzweig: The federation fallacy

Posted Mar 4, 2019 19:45 UTC (Mon) by roc (subscriber, #30627)
Parent article: Rosenzweig: The federation fallacy

The author seems to argue that the Web is not an example of a successfully decentralized system because extremely large companies that don't respect user privacy are able to operate on the Web and attract a large fraction of Web traffic.

This sets the bar for success unreasonably high. It means a successful decentralized system must somehow enforce norms on its participants in spheres well beyond the scope of the system itself. It also suggests a good system will somehow prevent any one participant for attracting "too much" traffic. Both of those features would imply restrictions on freedom, which would probably make people uncomfortable when you have worked out the details.

In fact the Web is a great example of a decentralized system. Contra the author, it *is* easy for anyone technical to set up a Web server and in practice a lot of people and organizations do set up their own servers and attract large amounts of traffic. (There are also numerous services to support non-technical people publishing Web content.) The standards that govern the technology are not controlled by any one company (as long as Mozilla survives). You have lots of viable choices for technology and services to set up a Web site, most of which are open-source. You also have multiple viable independent open-source clients. Web standards support delivering almost any kind of application and content. Sweeping all this decentralized goodness aside because Facebook and Youtube exist seems short-sighted.

home server prohibition matters i think

Posted Mar 4, 2019 21:32 UTC (Mon) by Garak (guest, #99377) [Link] (3 responses)

+1 generally, but I'll reiterate my pet theory here-

I think there is a general conspiracy[*] to tilt the playing field against the home server utilizer. I think this conspiracy profits those pursuing the centralized model. I think that if home server prohibition had been addressed by the net neutrality proponents (other than me and a seemingly very few others), I believe that we would have seen things like squirrelmail evolve into dramatically more appealing solutions than gmail. I haven't read the full article yet, but I hope Rosenzweig mentioned the issue of lowest-cost-tier common ISP home server prohibition. And again, it doesn't matter if the ISP doesn't even enforce it, as long as it is in the ToS it IMO radically shifts the motivational dynamic for home server software developers to the point that home server software of viable quality does not get developed in significant enough quantity to be more clearly relevant to the masses.

Though generally +1 again, reiterating that what's more important than decentralization-sans-behemoths is a decentralizable *option/platform* available to all. It's just like Free Speech generally. The important thing isn't that everybody is churning out some steady amount of Free Speech. The important thing is that everybody *COULD IF THEY WANTED TO* (without being taxed by some unnecessary thug/advertiser middleperson/serveroperator).

[*]
https://arstechnica.com/information-technology/2013/07/google-we-can-ban-servers-on-fiber-without-violating-net-neutrality/
https://arstechnica.com/information-technology/2013/10/google-fiber-now-explicitly-permits-home-servers/
https://ecfsapi.fcc.gov/file/7522219498.pdf
http://cloudsession.com/dawg/downloads/misc/kag-draft-2k121024.pdf
http://cloudsession.com/dawg/downloads/misc/kag-draft-2k121024.txt

home server prohibition matters i think

Posted Mar 4, 2019 21:40 UTC (Mon) by roc (subscriber, #30627) [Link] (2 responses)

> The important thing is that everybody *COULD IF THEY WANTED TO* (without being taxed by some unnecessary thug/advertiser middleperson/serveroperator).

Right. In practice it's also important that significant numbers of people *do* continually exercise the option, otherwise the systems that support the option "in theory" will atrophy. The Web does have that critical mass.

home server prohibition matters i think

Posted Mar 4, 2019 22:52 UTC (Mon) by Garak (guest, #99377) [Link] (1 responses)

And my thesis is that if the FCC had said "wow, yeah, that's an obvious problem with network neutrality, we need to give these ISPs a basic education on the Internet Protocol's every node can be a server, client, or both functionality" then we would have seen a critical mass on the (FOSS and commercial) home server software development and userbase side that is conspicuously absent, as evidenced by the present article/discussion. But by accepting the status-quo of predominant default home server prohibition terms of service, we have an environment where many average types get the same queesy feeling about running a home server as they do downloading a recent movie via bittorrent from 'thepiratebay' or the like. And that's all the motivational impediment you need to see the prospects of the home server software developer drying up to the point they sign an nda/noncompete and their soul over for a six figure salary to one of the major profiteers of the conspiracy. Bottom line, there are some big player$ that like things as they are, and they aren't going to be disrupted without a well financed fight from their side.

home server prohibition matters i think

Posted Mar 19, 2019 10:30 UTC (Tue) by nix (subscriber, #2304) [Link]

Yeah right. You can tell because nobody uses Google in the UK, or Facebook. The UK has *more stringent* rules than you propose (or, in fact, it has rules that produce extensive ISP competition, and some of those ISPs not only allow but encourage you to run home servers), but I'm writing this on Chromium and I use Google for searching almost to the exclusion of all else: indeed Google has more of a stranglehold here than in the US.

I am in favour of home servers -- I run quite a lot of them -- but all your harping on the subject is doing here is proving your parochialism. Your proposed solution has been tried and it *does not work*. It does not do what you repeatedly say it will do. I wish it did, but it turns out that allowing home servers is not a panacea. It doesn't cure disease or old age either, imagine that.

Rosenzweig: The federation fallacy

Posted Mar 4, 2019 22:05 UTC (Mon) by ejr (subscriber, #51652) [Link] (8 responses)

The web is nearly read-only.

Conversations and frequent interactions on small items are rather different.

Rosenzweig: The federation fallacy

Posted Mar 5, 2019 2:43 UTC (Tue) by areilly (subscriber, #87829) [Link] (1 responses)

What do you mean by that comment, in the light of the existence of that comment, on the web? Care to elaborate?
Are you pointing out that you can't just comment on or modify or add-to most other people's web sites (duh), or that it somehow isn't possible to publish your own content (write) on the web in general?

(federatable) web overlay commentary/etc

Posted Mar 10, 2019 7:19 UTC (Sun) by Garak (guest, #99377) [Link]

Are you pointing out that you can't just comment on or modify or add-to most other people's web sites (duh),

Actually that sounds a lot like something I recall g+ having deployed a few years back. The ability to +1 arbitrary pages, maybe comment as well. Of course the non-evil way to go about accomplishing that would be to utilize a federated network of such overlay content stored locally or otherwise under the netizen's control. I.e. no need for a centralized (commercial) big player to have control over the data involved (and utilizing every means possible to extract as much profit from access to the data as well as dominating/influencing the implementation details)

Rosenzweig: The federation fallacy

Posted Mar 5, 2019 4:53 UTC (Tue) by flussence (guest, #85566) [Link]

The web has long-since ceased to be read-only. It's r-x or rwx nowadays, you don't get very far without surrendering to arbitrary code execution.

Rosenzweig: The federation fallacy

Posted Mar 7, 2019 9:05 UTC (Thu) by callegar (guest, #16148) [Link] (4 responses)

Indeed, it is not by chance that the internet connection you get at home is Asymmetric (the A in ADSL), with a huge bandwidth to read and a relatively poor one to write.

This is also a significant disincentive for providing services from the home.

Cloud services, where many people need to upload files onto the cloud and get bored about waiting for the upload may lead ISPs to slightly reduce the asymmetry, though.

up/down asymmetry engineering, terms of service, disincentives, supply and demand

Posted Mar 10, 2019 7:09 UTC (Sun) by Garak (guest, #99377) [Link] (3 responses)

I would guess the ISPs would claim they've engineered/tuned the infrastructure to best meet supply and demand. I would immediately follow that up however with the idea that the causality chain starts with the common home server prohibition ToS, leading to demand characteristics, that the ISP can then use to justify tuning their network as they do. But while I believe in the long run that asymmetry factor will decrease, I think there is so much that can be done even as things are now that it's just ridiculous(ly sad how the profiteers get away with rigging the system in that way)

up/down asymmetry engineering, terms of service, disincentives, supply and demand

Posted Mar 10, 2019 14:34 UTC (Sun) by mpr22 (subscriber, #60784) [Link] (2 responses)

I have no particular interest in providing videos or video games with bulky audiovisual assets from my third-floor council flat, but I have plenty of interest in consuming those things, so even with a home server, I'd still have heavily asymmetric bandwidth needs.

television 3.0

Posted Mar 11, 2019 0:08 UTC (Mon) by Garak (guest, #99377) [Link] (1 responses)

true enough. Perhaps the more important point is that there are less well known dynamics of that engineering/tuning. I.e. if the cable modem provider is offering 25Mbps down and 3Mbps up it is not simply their choice to retune to 14 up and 14 down IIRC. And if home servers were protected by a network neutrality law/policy it wouldn't take such a 50/50 balance to accomplish many usual high bandwidth things. I.e. distributed(bittorrentesque) streaming/distribution is already widely succussfully used as an alternative to paying a CDN to help mitigate the source bandwidth bottleneck issue. While of course many bittorrent peers(cough simultaneous client and server behavior cough) are not home servers with access to greater outbound bandwidth, even if you had to have 10 or 1000 100kbps peers/cdn-amplifier-nodes/homelinuxservers, the method is clearly viable for the key example of an alternative fully decentralized video distribution network able to accomplish the same things as broadcast and cable television networks.

television 3.0

Posted Mar 11, 2019 10:24 UTC (Mon) by farnz (subscriber, #17727) [Link]

In large part, though, that's because the US hasn't built home Internet infrastructure; they have repurposed infrastructure designed for television (cable, DSL) for home Internet service, instead of putting in dedicated networking facilities.

This makes offering service very cheap - most of the civils have been done already in order to provide subscription TV (cable) or telephone networks (DSL - which was designed to let telcos compete with cable networks by offering TV), but also means that the compromises that make sense for TV (limited bandwidth from home to central office, much wider bandwidth from central office to home, more control of signal at central office thus higher modulation rates possible getting more bits/symbol) have to be accepted in terms of Internet access.

Fixing that requires fresh civils that replace the existing last mile networks with either dedicated copper or fibre (probably fibre nowadays, as it's cheaper in the volumes that a new network would need, and has far higher bandwidth in each direction than expensive copper - expensive copper can be good to around 5 GHz at best, but has attenuation on the order of 60 dB/km, while single mode fibre is good for around 100 THz - 100,000 GHz - with attenuation on the order of 1 dB/km).

This, in turn, requires either political willpower to spend tax money on disruptive infrastructure projects, or commercial incentives to do so rather than just providing Internet access on existing (paid-for) infrastructure. It's worth noting that in many former Soviet countries, where TV and telephone infrastructure did not exist, they're doing just that; putting in cheap fibre and running symmetric Internet services on it, because it's cheaper to do that than put in US-style TV and telephone infrastructure.

Similarly, parts of Scandinavia, Singapore, and South Korea are putting in fibre for Internet service because the political willpower is there to say "we want good Internet service, and we'll pay the price to get there, bypassing Internet over legacy installs.

Finally, in countries like the UK, there's a different route being tried to make it work commercially; we're doing fibre-to-the-cabinet (in the form of HFC cable and VDSL2 from telephone cabinets), which effectively moves the central offices closer to people's homes, and reduces the cost of replacing the old TV/telephone network with a pure fibre data-first network by making money from moving the switch to Internet services closer to people's homes. It's a lot cheaper to replace the ~300m of cable from my house to the nearest cabinet than it is to replace the ~5km of cable from my house to the central office.